Problem Solving for Scientific Computing: Data Modelling instead of Algorithms?
Victor G. Ganzha · mediaTUM (Technical University of Munich) · 2018
We present an approach to rapid solving of modelling problems for fluid dynamics. We show how the problem solving process in fluid dynamics can be expressed through the incremental successive data modelling on the basis of graphical description techniques like class diagrams in UML. As will be shown, in contrast to algorithmic approach the data modelling in terms of objects and relations among them enable one to describe the physical and mathematical properties and relations of the fluid flow. At the same time, it provides the necessary facilities to obtain the numerical code automatically from an abstract problem description. Moreover the tool prototype called GROOME to numerical code generation on the basis of graphical object oriented description techniques is presented. In order to demonstrate these facilities we consider a fluid dynamics problem involving shock waves in two-dimensional inviscid gas flow.